Jordan Rose 82ae9898ef [analyzer] Treat all struct values as regions (even rvalues).
This allows us to correctly symbolicate the fields of structs returned by
value, as well as get the proper 'this' value for when methods are called
on structs returned by value.

This does require a moderately ugly hack in the StoreManager: if we assign
a "struct value" to a struct region, that now appears as a Loc value being
bound to a region of struct type. We handle this by simply "dereferencing"
the struct value region, which should create a LazyCompoundVal.

This should fix recent crashes analyzing LLVM and on our internal buildbot.

<rdar://problem/12137950>

llvm-svn: 163066
2012-09-01 17:39:09 +00:00

156 lines
3.3 KiB
C++

// RUN: %clang_cc1 -analyze -analyzer-checker=core,alpha.core,debug.ExprInspection -analyzer-store=region -analyzer-constraints=range -verify -Wno-null-dereference %s
void clang_analyzer_eval(bool);
typedef typeof(sizeof(int)) size_t;
void malloc (size_t);
void f1() {
int const &i = 3;
int b = i;
int *p = 0;
if (b != 3)
*p = 1; // no-warning
}
char* ptr();
char& ref();
// These next two tests just shouldn't crash.
char t1 () {
ref() = 'c';
return '0';
}
// just a sanity test, the same behavior as t1()
char t2 () {
*ptr() = 'c';
return '0';
}
// Each of the tests below is repeated with pointers as well as references.
// This is mostly a sanity check, but then again, both should work!
char t3 () {
char& r = ref();
r = 'c'; // no-warning
if (r) return r;
return *(char*)0; // no-warning
}
char t4 () {
char* p = ptr();
*p = 'c'; // no-warning
if (*p) return *p;
return *(char*)0; // no-warning
}
char t5 (char& r) {
r = 'c'; // no-warning
if (r) return r;
return *(char*)0; // no-warning
}
char t6 (char* p) {
*p = 'c'; // no-warning
if (*p) return *p;
return *(char*)0; // no-warning
}
// PR13440 / <rdar://problem/11977113>
// Test that the array-to-pointer decay works for array references as well.
// More generally, when we want an lvalue for a reference field, we still need
// to do one level of load.
namespace PR13440 {
typedef int T[1];
struct S {
T &x;
int *m() { return x; }
};
struct S2 {
int (&x)[1];
int *m() { return x; }
};
void test() {
int a[1];
S s = { a };
S2 s2 = { a };
if (s.x != a) return;
if (s2.x != a) return;
a[0] = 42;
clang_analyzer_eval(s.x[0] == 42); // expected-warning{{TRUE}}
clang_analyzer_eval(s2.x[0] == 42); // expected-warning{{TRUE}}
}
}
void testNullReference() {
int *x = 0;
int &y = *x; // expected-warning{{Dereference of null pointer}}
y = 5;
}
void testRetroactiveNullReference(int *x) {
// According to the C++ standard, there is no such thing as a
// "null reference". So the 'if' statement ought to be dead code.
// However, Clang (and other compilers) don't actually check that a pointer
// value is non-null in the implementation of references, so it is possible
// to produce a supposed "null reference" at runtime. The analyzer shoeuld
// still warn when it can prove such errors.
int &y = *x;
if (x != 0)
return;
y = 5; // expected-warning{{Dereference of null pointer}}
}
void testReferenceAddress(int &x) {
clang_analyzer_eval(&x != 0); // expected-warning{{TRUE}}
clang_analyzer_eval(&ref() != 0); // expected-warning{{TRUE}}
struct S { int &x; };
extern S getS();
clang_analyzer_eval(&getS().x != 0); // expected-warning{{TRUE}}
extern S *getSP();
clang_analyzer_eval(&getSP()->x != 0); // expected-warning{{TRUE}}
}
void testFunctionPointerReturn(void *opaque) {
typedef int &(*RefFn)();
RefFn getRef = (RefFn)opaque;
// Don't crash writing to or reading from this reference.
int &x = getRef();
x = 42;
clang_analyzer_eval(x == 42); // expected-warning{{TRUE}}
}
// ------------------------------------
// False negatives
// ------------------------------------
namespace rdar11212286 {
class B{};
B test() {
B *x = 0;
return *x; // should warn here!
}
B &testRef() {
B *x = 0;
return *x; // should warn here!
}
}